Functional Connectivity Between the Posterior Default Mode Network and Parahippocampal Gyrus Is Disrupted in Older Adults with Subjective Cognitive Decline and Correlates with Subjective Memory Ability

Abstract

Background:

Subjective cognitive decline (SCD) is associated with increased risk of developing Alzheimer’s disease (AD). However, the underlying mechanisms for this association remain unclear. Neuroimaging studies suggest the earliest AD-related changes are large-scale network disruptions, beginning in the posterior default mode (pDMN) network.

Objective:

To examine the association between SCD and pDMN network connectivity with medial temporal lobe (MTL) regions using resting-state functional magnetic resonance imaging.

Methods:

Forty-nine participants with either SCD (n = 23, 12 females; mean age: 70.7 (5.5)) or who were cognitively unimpaired (CU; n = 26, 16 females, mean age: 71.42 (7.3)) completed the Memory Functioning Questionnaire, a measure of subjective memory, and underwent resting state functional MRI at 3 Tesla. Functional connectivity between the posterior cingulate cortex (PCC), as the key pDMN node, and MTL regions were compared between SCD and CU groups. Further, the association between pDMN-MTL connectivity and the Frequency of Forgetting subscale of the Memory Functioning Questionnaire was examined.

Results:

Connectivity between the PCC-MTL was observed in the CU group but was absent in SCD (t(47) = 2.69, p = 0.01). Across all participants, self-perception of frequency of forgetting, but not objective memory, was strongly correlated with connectivity between the PCC-left parahippocampal gyrus (r = 0.43, p = 0.002).

Conclusion:

These findings support the hypothesis that increased AD risk in SCD may be mediated by disrupted pDMN-parahippocampal connectivity. In addition, these findings suggest that frequency of forgetting may serve as a potential biomarker of SCD due to incipient AD.

Keywords

Alzheimer’s disease default mode network depression functional connectivity memory functioning questionnaire parahippocampal gyrus posterior cingulate cortex resting state functional magnetic resonance imaging subjective cognitive decline subjective memory

INTRODUCTION

Subjective cognitive decline (SCD) is defined by the presence of concerns regarding cognitive changes such as memory decline, but intact performance on formal testing [1]. Despite the absence of objective cognitive impairment, SCD is conceptualized as a state of increased risk for Alzheimer’s disease, because individuals with SCD are more likely to show evidence of Alzheimer’s disease biomarkers and, when followed longitudinally, they are more likely to develop future Alzheimer’s disease compared to those without subjective cognitive concerns [1, 2]. However, the neural mechanisms linking SCD with Alzheimer’s disease risk remain unclear.

Although predominant pathophysiological models of Alzheimer’s disease have focused on neuropathological changes in the brain, including the appearance of amyloid plaques and neurofibrillary tangles, neuroimaging findings suggest that large-scale network disruptions are among the earliest changes in Alzheimer’s disease [3]. Using multimodal neuroimaging and clinical phenotyping data to characterize the pattern of default mode network (DMN) subsystem connectivity changes across the entire Alzheimer’s disease spectrum, Jones and colleagues [3] described a systems-level model of Alzheimer’s disease in which the posterior DMN (pDMN) fails prior to imaging evidence of neuropathological changes, such as amyloid deposition, in the brain. Their model suggested that the failure of the pDMN initiates a connectivity cascade that progresses throughout the Alzheimer’s disease spectrum, which may lead to compensatory changes characterized by increased connectivity between the pDMN and hubs within the prefrontal cortex and other regions in the DMN.

The DMN is a set of functionally connected brain regions, with ventral medial prefrontal cortex and retrospenial/posterior cingulate cortex (PCC) as key nodes of the anterior and pDMN respectively, and other regions including the inferior parietal lobule, dorsal medial prefrontal cortex, and hippocampal formation, consisting of the entorhinal cortex and parahippocampal gyrus (PHG) [4, 5]. Initial descriptions conceptualized the DMN as a system active during rest, or in mental states during which cognitive demands are minimal such as mind wandering, but subsequent work has also implicated the DMN in memory processes [4]. Using task-based functional magnetic resonance imaging (fMRI), Andrews-Hanna et al. 2010 [6] demonstrated that nodes within the DMN could be functionally segregated into three subsystems that included a midline core system containing both the PCC and anterior medial prefrontal cortex, which was activated during construction of mental scenes based on memory, a subcortical medial temporal lobe system, and a dorsal medial prefrontal cortex system that was engaged in self-referential processing [6].

Alterations in connectivity within the DMN in Alzheimer’s disease differ among the subsystem regions, with decreased connectivity found in PCC and medial temporal lobe subsystems and increased connectivity observed in medial prefrontal cortex regions [7]. Findings regarding other regions within the DMN are highly variable, as suggested by a recent meta-analysis of neuroimaging studies in the Alzheimer’s disease prodrome of mild cognitive impairment (MCI), that concluded that reduced PCC connectivity was the largest and most significant group effect, even while this was seen in just over half of studies [8]. Previous studies in SCD similarly show reduced connectivity between pDMN and hippocampus or PHG [9 –11], although increased connectivity between pDMN and a network of medial temporal lobe, parietal and ventral medial prefrontal cortex regions was reported in one study [12]. A study using resting-state functional magnetic resonance imaging (rs-fMRI) found that cognitive complaint scores in older adults with normal cognition were inversely correlated with functional connectivity of resting-state networks but did not specifically examine pDMN connectivity [13].

In the current study, we used rs-fMRI to compare pDMN connectivity between SCD and cognitively unimpaired (CU) older adults [14], and to examine the relationship between pDMN connectivity and subjective memory ability. Based on the extant literature, we made the following hypotheses: 1) SCD participants would show decreased connectivity between the pDMN and medial temporal regions compared to CU and 2) among older adults with normal neuropsychological test performance, subjective memory, but not objective memory performance, would correlate with connectivity between the pDMN and medial temporal regions, including both the hippocampus and PHG.

MATERIALS AND METHODS

Participants

Forty-nine older adults who were classified either as SCD [n = 23; 12 females; mean age 70.7(SD5.5)] or CU [n = 26; 16 females, mean age 71.42 (SD7.3)] based on normal neuropsychological test performance and the presence or absence of subjective memory concerns respectively, participated in the study. SCD was ascertained by affirmative responses to two questions on a self-report questionnaire: “Do you feel your memory is becoming worse?” “If so, are you concerned?”[2]. Participants were recruited from Baycrest’s mood or memory clinics, the research volunteer database at Rotman Research Institute, or through advertising in the community. Eligibility criteria included 60 years of age or older, English language proficiency, performance within the normal based on age, gender and education-adjusted norms on a neuropsychological battery, no neurological disorders, no active or past/lifetime history of psychiatric conditions, and no use of psychotropic medications within the preceding 3 weeks. The study was approved by the Research Ethics Board of Baycrest.

Measures

Memory functioning questionnaire (MFQ)

The MFQ is a validated self-report measure of subjective memory ability that has been used in healthy older adults and clinical samples [15, 16]. MFQ items cluster into four subscales: Frequency of Forgetting (MFQ-FF), Seriousness of Forgetting, Retrospective Functioning, and Mnemonics Usage with higher scores on the MFQ indicating better subjective memory ability. The MFQ-FF subscale has been shown to be associated with increased in vivo cerebral amyloid pathology on positron emission tomography (PET) imaging in CU or MCI [17 –19], suggesting that the MFQ-FF subscale may provide an index of SCD due to incipient Alzheimer’s disease. Thus, MFQ-FF was selected as a measure of subjective memory ability.

Montreal cognitive assessment (MoCA)

The MoCA is a global measure of cognition with greater sensitivity to detect impaired executive function in older adult populations [20] that was included as a covariate in analyses to control for variability in overall cognition.

Geriatric depression scale (GDS)

The GDS is a self-report scale of depression for use in geriatric populations [21]. Because subjective cognitive concerns are strongly linked to depression [22, 23], the GDS was included as a covariate in analyses to control variability in depressive symptomatology.

Neuropsychological assessment

A complete neuropsychological test battery was administered to all participants to establish cognitive status. A summary of test scores is presented in Table 1. The battery included the Wechsler Abbreviated Intelligence Scale III Vocabulary Test, 15-item Boston Naming Test, Delis-Kaplan Executive Function System Phonemic Fluency, Category Fluency, and Trail Making Tests, Weschler Memory Scale-Revised Logical Memory, and the California Verbal Learning Test-II (CVLT-II).

Table 1

Demographic and clinical characteristics of Cognitively Unimpaired and Subjective Cognitive Decline groups

Variable	Cognitively Unimpaired (n = 26)	Subjective Cognitive Decline (n = 23)
Age at baseline, y	71.42 (7.3)	70.70 (5.5)
Number of women in sample	16	12
Education, y	16.38 (3.0)	17.43 (2.5)
Memory Functioning Questionnaire
Total Score^β	336 (48.7)	293.7 (58.4)
Frequency of Forgetting score^β	191.8 (27.1)	168.5 (34.1)
Montreal Cognitive Assessment	27.35 (1.7)	26.91 (1.6)
Geriatric Depression Scale^¥	.62 (.90)	1.52 (1.7)
Wechsler Abbreviated Scale of Intelligence Matrix Reasoning	20.08 (5.3)	20.17 (3.8)
Wechsler Adult Intelligence Scale–III Vocabulary	47.08 (5.9)	46.57 (4.8)
Boston Naming Test (15-item)	37.04 (21.9)	30.70 (20.7)
Delis-Kaplan Executive Function System
Phonemic Fluency	49.12 (11.1)	45.83 (10.3)
Category Fluency	43.44 (12.3)^*	39.26 (6.2)
Trail Making Test Number-Letter Switching Scaled Score	12.35 (3.1)	12.09 (2.2)
Wechsler Memory Scale-Revised
Logical Memory Immediate Recall	14.0 (4.0)	13.04 (2.8)
Logical Memory Delayed Recall	12.58 (3.7)	11.13 (3.4)
California Verbal Learning Test-II
List A Learning (Trials 1–5)^β	54.19 (9.6)	47.87 (10.5)
List A Short Delayed Free Recall^β	11.46 (2.8)	9.52 (2.9)
List A Long Delayed Free Recall^β	12.35 (2.6)	10.09 (3.1)

Summary statistics are mean (SD) for continuous measures. ^βSCD < CU, p < 0.05; ^¥SCD > CU, p < 0.05; see text for statistics. ^*data missing for one participant.

The CVLT-II is a standard measure of verbal memory based on wordlist learning [24]. Long delayed free recall on the CVLT-II (CVLT-II-delayed-recall) is predictive of progression to dementia [23, 25] and is a component of the Alzheimer Disease Cooperative Study Preclinical Alzheimer Cognitive Composite [26]. Since CVLT-II-delayed-recall is sensitive to objective memory decline, it was included in analyses of the relationship between pDMN-medial temporal cortex connectivity and memory.

Magnetic resonance imaging (MRI) acquisition

Participants were scanned at 3T (Siemens TIM Trio, Erlangen, Germany) using a 12-channel array head coil. A 3D T1-weighted anatomical scan was acquired using magnetization prepared rapid gradient echo (Magnetization Prepared: Rapid Gradient Echo; parameters 1×1×1 resolution, TR =2000 ms, TI = 1100 ms, TE = 2.22 ms, flip angle = 9°, FOV = 256×256 mm, matrix = 256×256,176 slices (ascending order), GRAPPA factor = 2, oblique sagittal orientation). A 6 min rs-fMRI scan, in an awake state with eyes closed, was acquired using gradient-echo echo-planar imaging blood oxygen level-dependent sequence (voxel size = 3.1×3.1×5 mm³, TR = 2000 ms, TE = 30 ms, flip angle = 70°, FOV = 200×200 mm, matrix = 64×64, 30 slices, 5 mm thickness, oblique axial orientation).

Preprocessing of MRI data

Structural MRI

T1-weighted images were processed using FreeSurfer (v 6.0; available at http://surfer.nmr.mgh.harvard.edu) which includes cortical reconstruction and volumetric segmentation [27]. Automated segmentation of white matter, grey matter, cerebrospinal fluid, and estimated total intracranial volume based on probabilistic information was extracted and used for correction in functional segmentation. No parallel imaging was used.

Functional MRI

fMRI data were preprocessed using FreeSurfer Functional Analysis Stream [28] using the following steps: 1) intensity normalization, 2) registration to structural images using reconstructed cortical surfaces, 3) motion correction using six degrees of freedom [29], 4) physiological motion correction, 5) slice-timing correction, 6) re-sampling into Montreal Neurological Imaging 152 space, and 7) smoothing with Gaussian kernel of 5 mm³ at full-width half maximum [30]. Preprocessed functional images were imported into the CONN toolbox [31] with SPM12 for further correction. Outliers were identified using the Artifact Detection Tool [32]. BOLD signal data were passed through a temporal band pass filter (0.009–0.08 Hz) after regression of white matter and cerebrospinal fluid. The mean BOLD signal time course was extracted from regions-of-interest (ROI) and correlated with the time course for each source ROI. No parallel imaging was used.

Data analysis

ROIs

The PCC, a key node within the pDMN, was selected as the seed ROI and was defined manually in Montreal Neurological Imaging space as an 8 mm diameter sphere centered on x = –6, y = –52, z = 40 [33, 34]. Source ROIs were other DMN regions including medial prefrontal cortex, and bilateral hippocampus, anterior and posterior PHG, and lateral parietal cortex, as defined by the CONN toolbox using the FSL Harvard-Oxford atlas.

Connectivity analyses

A second-level general linear model was created to compute connectivity matrices between each pair of ROIs, separating between-subject and between-source contrasts, to obtain population level-estimates for CU and SCD groups. The threshold for statistical significance was set at the false-discovery rate-corrected p value of 0.05.

Correlational analyses

The relationship between subjective memory and pDMN-medial temporal cortex connectivity was analyzed within the entire sample since both SCD and CU participants had normal neuropsychological test performance. The medial temporal regions included left and right hippocampus and left and right anterior and posterior PHG (six ROIs total). Fisher-transformed Z-scores of bivariate correlations were calculated between the PCC and each medial temporal region to yield correlation coefficients as measures of functional connectivity for each participant. These values were entered into correlational analyses to examine the association between PCC-medial temporal cortex functional connectivity and both subjective and objective memory measures (MFQ-FF, CVLT-II-delayed-recall) with statistical significance set at a Bonferroni-corrected threshold of p < 0.004 [0.05/(six ROIs x two memory measures) = 0.05/12 comparisons = 0.004]. Statistical Program for the Social Sciences for Windows version 26 was used to perform correlational analyses.

RESULTS

Sample characteristics

Table 1 summarizes demographic and clinical characteristics of CU and SCD participants. Age, gender composition, and educational level were similar in both groups. MFQ-FF scores were lower in SCD compared to CU participants (t (47) = 2.67, p =0.01). All forty-nine participants had normal neuropsychological performance. However, despite normal neuropsychological test performance, CVLT-II memory scores were lower in SCD, relative to CU (CVLT-II immediate recall: t(47) = 2.20, p =0.03; CVLT-II-short delayed free recall: t(47) = 2.38, p = 0.02; CVLT-II-long delayed free recall: t(47) =2.78, p = 0.008). In addition, although all participants scored within normal on the GDS, scores were higher in the SCD group compared to CU (t(47) = 2.37, p = 0.02).

Group differences in pDMN connectivity

Functional connectivity maps of statistically significant correlations between the PCC and ROIs within the DMN for both CU and SCD groups are shown in Fig. 1. As summarized in Fig. 1 and Table 2, within the CU group, the PCC was functionally connected to the medial prefrontal cortex and to bilateral lateral parietal cortex, hippocampus, and posterior, but not anterior, PHG. In contrast, within the SCD group, the PCC was functionally connected to the medial prefrontal cortex and bilateral lateral parietal cortex only, with no connectivity between the PCC and any medial temporal lobe region. Functional connectivity between the PCC and left posterior PHG, but not with any other target region, was significantly different between SCD and CU [t (47) = 2.69, p = 0.01].

Fig. 1

Connectivity between the posterior cingulate cortex (PCC) and regions within the default mode network, significant at the false discovery rate corrected p-value of 0.05, in a) cognitively unimpaired and b) subjective cognitive decline. LPC, lateral parietal cortex; HC, hippocampus; mPFC, medial prefrontal cortex; PHG, parahippocampal gyrus.

Table 2

Functional connectivity between posterior cingulate cortex and regions-of-interest within the default mode network in Cognitively Unimpaired and Subjective Cognitive Decline groups

Regions -of-interest	Cognitively Unimpaired			Subjective Cognitive Decline
	beta	T(25)	p-FDR (seed-level correction)	beta	T(25)	p-FDR (seed-level correction)
Medial Prefrontal Cortex	0.15	3.22	0.016193^*	0.13	3.20	0.024916^*
Lateral Parietal Cortex
Left	0.20	5.12	0.000257^***	0.30	5.98	0.000144^***
Right	0.19	3.64	0.007453^**	0.11	2.91	0.040547^*
Hippocampus
Left	0.12	4.36	0.001524^***	0.02	0.44	0.815949
Right	0.13	4.07	0.002816^**	0.04	0.91	0.562971
Anterior Parahippocampal Gyrus
Left	0.03	1.11	0.411957	0.02	0.58	0.756616
Right	0.01	0.45	0.737509	0.00	0.08	0.967309
Posterior Parahippocampal Gyrus
Left^a	0.15	5.07	0.000277^***	0.04	1.14	0.480299
Right	0.11	3.02	0.024062^*	0.06	1.23	0.443325

^***represents significant ROI-ROI connection (p-FDR seed-level corrected < 0.001), ^**represents significant ROI-ROI connection (p-FDR seed-level corrected < 0.01), ^*represents significant ROI-ROI connection (p-FDR seed-level corrected < 0.05). FDR, False Discovery Rate. ^aSCD < CU, t (47) = 2.69, p = 0.01.

Association between PCC-medial temporal cortex connectivity and memory

Results of correlational analyses are summarized in Table 3. Across all participants, a significant association was found between MFQ-FF and functional connectivity between the PCC and left posterior PHG (r = 0.43, p = 0.002; Fig. 2), which remained statistically significant after controlling for GDS and MoCA (partial correlation = 0.403, p = 0.005). A positive correlation was also found between MFQ-FF and PCC-left hippocampal connectivity, but this was not significant at the Bonferroni-corrected threshold (r = 0.30, p = 0.04).

Table 3

Association between posterior default mode network –medial temporal lobe functional connectivity and subjective and objective memory in older adults with normal neuropsychological test performance (N = 49)

Region of Interest	Memory Functioning Questionnaire–Frequency of Forgetting	California Verbal Learning Test-II Long Delayed Free Recall
Hippocampus
Left	0.30 (0.04)	0.014 (0.92)
Right	0.231 (0.1)	0.007 (0.96)
Anterior Parahippocampal Gyrus
Left	–0.086 (0.6)	0.010 (0.95)
Right	–0.012 (0.9)	0.008 (0.96)
Posterior Parahippocampal Gyrus
Left	0.430 (0.002)^*	0.249 (0.09)
Right	0.164 (0.3)	–0.025 (0.87)

Values in the table are Pearson’s r (p-value). ^*Significant at Bonferroni-adjusted threshold of p < 0.004.

Fig. 2

CU, Cognitively unimpaired; SCD, Subjective cognitive decline. Relationship between Memory Functioning Questionnaire - Frequency of Forgetting score and functional connectivity between the posterior cingulate cortex and left posterior parahippocampal gyrus in older adults with normal neuropsychological test performance based on age, gender, and education-adjusted norms. Note that the figure illustrates values for CU and SCD participants separately to demonstrate that the correlation is not driven by group differences in either variable. The correlation analysis was conducted in the entire sample of older adults (n = 49).

In contrast, CVLT-II-delayed-recall scores were not correlated with connectivity between PCC and any medial temporal region.

DISCUSSION

The main findings of the current study were a reduction in resting-state functional connectivity between the pDMN and medial temporal lobe in SCD, and evidence of an association between pDMN-medial temporal cortex connectivity and subjective, but not objective, memory ability in older adults with normal cognition based on neuropsychological test performance. Specifically, while CU participants showed significant connectivity between the PCC, a key node within the pDMN, and bilateral hippocampus and posterior PHG, the SCD group did not exhibit connectivity between PCC and any medial temporal region. In particular, the SCD group showed decreased connectivity between the pDMN and left posterior PHG compared to CU. Across all participants with normal cognition, subjective memory, based on MFQ-FF scores, was strongly correlated with connectivity between the pDMN and the left posterior PHG, but not hippocampus or any other medial temporal region. In other words, greater connectivity between the pDMN and the left posterior PHG was associated with self-perception of better memory. This relationship remained statistically significant when depression symptoms and overall cognition were accounted for using partial correlation analysis.

To our knowledge, this is the first report of a reduction in PCC-PHG connectivity in SCD and an association between PCC-PHG functional connectivity and subjective memory in older adults with normal cognition based on neuropsychological assessment. The association of pDMN and its connectivity with the PHG appeared to be specific to subjective aspects of memory, i.e., self-awareness of memory ability, because actual or objective memory performance, as quantified by delayed recall performance on the CVLT-II, was not correlated with pDMN-PHG connectivity. Our findings of a reduction in functional connectivity between the pDMN and both PHG and hippocampus in the SCD group extend existing evidence of alterations in resting-state functional connectivity in SCD or MCI [9–11 , 35–38]. Consistent with our findings, two previous studies of SCD that examined pDMN-medial temporal cortex functional connectivity using rs-fMRI show decreased connectivity between the PCC and left or right hippocampus, although neither defined the PHG as an ROI [9, 10]. Our data are also consistent with a magnetoencephalography (MEG) study in SCD that showed decreased connectivity between PHG and angular gyrus, a region within the inferior parietal lobule that is part of the pDMN [11].

In contrast, an rs-fMRI study of cognitively-normal older adults with a family history of Alzheimer’s disease reported increased connectivity between pDMN and a “medial temporal memory system” consisting of hippocampus, PHG, retrosplenial, inferior parietal and ventral medial prefrontal cortex [12]. However, these findings are difficult to interpret since their medial temporal memory system included the ventral medial prefrontal cortex, a key node within the anterior DMN, and other cortical regions in addition to medial temporal cortex. Since greater connectivity between anterior and posterior DMN, thought to be an early compensatory response, has been reported in older adults at risk for Alzheimer’s disease [3], whether the increased pDMN-medial temporal memory system connectivity can be attributed to greater connectivity between pDMN and medial temporal structures is unknown. In addition, methodological issues such as querying perception of cognitive decline, but not concern [1], and a gap of up to 1 year between neuropsychological assessment and ascertainment of SCD, may account for their study findings.

We note that the correlation between pDMN-PHG connectivity with subjective memory is consistent with our findings of a reduction in pDMN connectivity with the medial temporal cortex in SCD relative to CU participants. In the current study, older adults with normal neuropsychological test performance who had lower MFQ-FF scores, i.e., perception of poorer memory ability, also showed decreased connectivity between the PCC and left PHG. A similar, but weaker, association was observed for PCC-hippocampal connectivity. We are aware of only one other study which demonstrated an association between subjective concerns about memory and resting-state functional connectivity. Contreras and colleagues reported that cognitive complaint scores in older adults with normal cognition were inversely correlated with functional connectivity of resting-state networks [13]. However, this study did not specifically examine pDMN connectivity.

Although we also predicted a positive association between MFQ-FF and connectivity between PCC and hippocampus, only a weak, non-statistically significant, correlation was observed. Notably, current evidence from resting-state functional connectivity studies in humans is inconsistent with respect to which specific regions of the medial temporal lobe are connected to midline cortical regions of the DMN at rest. Specifically, it is unclear whether the hippocampus has direct connections with DMN or whether it interfaces with the DMN through the PHG [39, 40]. In a functional MRI study of associative memory, Ward and colleagues demonstrated that several hippocampal encoding regions lacked significant connectivity with cortical DMN nodes during resting-state [41]. Moreover, a mediation analysis showed that resting-state functional connectivity between hippocampus and PCC is indirect and mediated by the PHG.

Similarly, others have hypothesized that the PHG may serve as the primary hub between the DMN and medial temporal memory network [42]. Volumetric MRI studies also suggest that the PHG is preferentially associated with subtle, age-related decline [35] and is better able to discriminate among cognitively-normal older adults, MCI, and Alzheimer’s disease, as compared with the hippocampus [43]. Thus, our finding of a significant association between subjective memory and pDMN connectivity with PHG, but not hippocampus, may be considered consistent with previous work that suggests an indirect connection between the hippocampus and pDMN.

Although both anterior and posterior PHG regions are implicated in memory decline [42, 44], our findings support a significant association between subjective memory and functional connectivity between pDMN and only the posterior region of the PHG. The posterior PHG plays a key role in conveying both cortical input to the hippocampus, as well as the output from the medial temporal system to cortical regions [41]. Thus, disrupted PHG connectivity may lead to downstream effects, including alterations of hippocampal connectivity with the pDMN. The greater correlation with the posterior, as opposed to anterior, PHG is consistent with more recent literature indicating that the posterior PHG is preferentially associated with age-related memory loss, relative to the hippocampus and other regions of the PHG [35] and with self-awareness of memory loss [45].

As a brief comment on laterality of findings, it is noteworthy that pDMN connectivity to left, rather than right, lateralized regions within medial temporal cortex was correlated with subjective memory ability, supporting studies of Alzheimer’s disease progression that show greater hypometabolism, and greater extent and rate of atrophy within the left PHG compared to the right [42, 46]. Future studies are needed to identify potential laterality effects in the progression of neural network disruption in the earliest stages of Alzheimer’s disease.

Finally, the findings of a direct association between the MFQ-FF and pDMN-medial temporal cortex functional connectivity extend results from previous studies which reported that MFQ-FF scores were correlated with in vivo amyloid burden [17 –19]. Taken together, the current and previous findings from the literature suggest that the MFQ-FF may be particularly sensitive to subjective memory changes associated with Alzheimer’s disease. Thus, while there is lack of consensus regarding the most appropriate measure of SCD due to incipient Alzheimer’s disease [47], the associations between MFQ-FF and Alzheimer’s disease-related neuroimaging biomarkers such as in vivo amyloid levels and our current findings of pDMN connectivity suggest the potential utility of the MFQ-FF as a means to discriminate between older adults with age-related subjective memory decline versus those with SCD that heralds incipient Alzheimer’s disease. This hypothesis needs to be tested in future studies.

A major strength of our study is that all participants were carefully assessed using psychiatric interview and neuropsychological assessment to establish cognitive status and to exclude current or history of psychiatric conditions that may contribute to memory changes. Past studies of SCD typically exclude only participants with current depression symptoms using the GDS, a screening tool that is not diagnostic for depression. Further, older adults with a past history of depression frequently show residual cognitive deficits, despite recovering from depression [48]. Although participants with acute or past depression were excluded from participation in the current study, we confirmed that the association between subjective memory ability and pDMN and PHG functional connectivity was not confounded by the presence of depression symptoms using partial correlation analyses.

Our study is not without limitations. The main limitation is that our sample consisted of older adults with SCD who may or may not develop Alzheimer’s disease. Thus, whether the findings in this study would be found in a preclinical Alzheimer’s disease sample as defined by amyloid positivity on PET imaging or cerebrospinal fluid analyses, is unclear. The second main limitation is the small sample size. Our sample size did not allow us to examine the effects of sex on our findings, an important consideration given that there are established sex differences in episodic memory, as well as risk for Alzheimer’s disease and Alzheimer’s disease prognosis [49]. In addition, we did not explore the role of genetic biomarkers and their impact on the link between SCD and future Alzheimer’s disease development.

Finally, it should be noted that although SCD participants scored within normal on all neuropsychological tests, including within the memory domain, their CVLT-II scores were significantly lower than the CU group. Similarly, while SCD participants scored within normal on the GDS, their depression scores were significantly higher than CU. This pattern of findings suggests an interplay among SCD, depression, and subtle memory decline that may contribute to, or reflect, pDMN-PHG connectivity. In a post-hoc exploratory analysis of the total sample, GDS scores were inversely correlated with connectivity between the pDMN and right PHG/hippocampal regions (threshold (r = –0.29, p = 0.045; r = –0.36, p = 0.012, respectively), a finding that is consistent with evidence of decreased pDMN connectivity with the right hippocampus in individuals who have recovered from depression [50].

We propose a model of preclinical AD in which the early disruption in pDMN connectivity [3] to the left medial temporal region manifests behaviorally as self-awareness of impaired encoding and retrieval processes that have not yet impacted on memory function, possibly due to compensatory mechanisms, while reduced pDMN connectivity to the right medial temporal lobe region is expressed clinically as mood dysregulation or depression. This model is compatible with the well-established associations between neuropsychiatric symptoms and AD risk [51 –53], and evidence of emotion dysregulation as potential risk marker of AD [54].

Notwithstanding limitations of the study, our findings of a reduction in pDMN and medial temporal cortex connectivity in SCD, and evidence of a link between self-perception of memory ability and strength of pDMN-PHG connectivity indicate that it may be possible to identify phenotypic features of large-scale network disruptions that are among the earliest pathophysiological changes in Alzheimer’s disease [3]. The present data suggest a potential neural mechanism for the association between SCD and Alzheimer’s disease risk and extend findings from previous PET studies which support the potential utility of the MFQ-FF as a measure of SCD due to incipient Alzheimer’s disease. In the absence of objective memory impairment, self-awareness of memory change, together with mood dysregulation, may be the first outward manifestation of neural network disruption in Alzheimer’s disease. Self-perception of memory decline may reflect increased effort in engaging encoding and retrieval processes to compensate for subtle memory dysfunction, and as such, SCD may mediate the association between functional connectivity and objective memory ability. These hypotheses warrant further investigations that incorporate additional memory measures, Alzheimer’s disease biomarkers and longitudinal follow-up to establish Alzheimer’s disease diagnoses in individuals with SCD.

Footnotes

ACKNOWLEDGMENTS

We thank Aliya Ali, Frankie Chan, and Carrie Shorey for assistance with data collection and Darren Liang for assistance with CONN analyses and contribution to initial drafts of the manuscript.

This work was funded by the Alzheimer Society of Canada, Alzheimer Society Research Program 16–03; Centre for Aging and Brain Health Innovation, Researcher-Clinician Partnership Program 2.

Authors’ disclosures available online ().

References

Jessen

, Amariglio

, Buckley

, van der Flier

, Han

, Molinuevo

, Rabin

, Rentz

, Rodriguez-Gomez

, Saykin

, Sikkes

SAM

, Smart

, Wolfsgruber

, Wagner

(2020) The characterisation of subjective cognitive decline. Lancet Neurol 19, 271–278.

Jessen

, Amariglio

, van Boxtel

, Breteler

, Ceccaldi

, Chetelat

, Dubois

, Dufouil

, Ellis

, van der Flier

, Glodzik

, van Harten

, de Leon

, McHugh

, Mielke

, Molinuevo

, Mosconi

, Osorio

, Perrotin

, Petersen

, Rabin

, Rami

, Reisberg

, Rentz

, Sachdev

, de la Sayette

, Saykin

, Scheltens

, Shulman

, Slavin

, Sperling

, Stewart

, Uspenskaya

, Vellas

, Visser

, Wagner

, Subjective Cognitive Decline Initiative Working Group (2014) A conceptual framework for research on subjective cognitive decline in preclinical Alzheimer’s disease. Alzheimers Dement 10, 844–852.

Jones

, Knopman

, Gunter

, Graff-Radford

, Vemuri

, Boeve

, Petersen

, Weiner

, Jack

Jr. , Alzheimer’s Disease Neuroimaging Initiative (2016) Cascading network failure across the Alzheimer’s disease spectrum. Brain 139, 547–562.

Raichle

, MacLeod

, Snyder

, Powers

, Gusnard

, Shulman

(2001) A default mode of brain function. Proc Natl Acad Sci U S A 98, 676–682.

Buckner

, Andrews-Hanna

, Schacter

(2008) The brain’s default network: Anatomy, function, and relevance to disease. Ann N Y Acad Sci 1124, 1–38.

Andrews-Hanna

, Reidler

, Sepulcre

, Poulin

, Buckner

(2010) Functional-anatomic fractionation of the brain’s default network. Neuron 65, 550–562.

Jones

, Machulda

, Vemuri

, McDade

, Zeng

, Senjem

, Gunter

, Przybelski

, Avula

, Knopman

, Boeve

, Petersen

, Jack

Jr. (2011) Age-related changes in the default mode network are more advanced in Alzheimer disease. Neurology 77, 1524–1531.

Eyler

, Elman

, Hatton

, Gough

, Mischel

, Hagler

, Franz

, Docherty

, Fennema-Notestine

, Gillespie

, Gustavson

, Lyons

, Neale

, Panizzon

, Dale

, Kremen

(2019) Resting state abnormalities of the default mode network in mild cognitive impairment: A systematic review and meta-analysis. J Alzheimers Dis 70, 107–120.

Dillen

KNH

, Jacobs

HIL

, Kukolja

, Richter

, von Reutern

, Onur

, Langen

, Fink

(2017) Functional disintegration of the default mode network in prodromal Alzheimer’s disease. J Alzheimers Dis 59, 169–187.

10.

Wang

, Risacher

, West

, McDonald

, Magee

, Farlow

, Gao

, O’Neill

, Saykin

(2013) Altered default mode network connectivity in older adults with cognitive complaints and amnestic mild cognitive impairment. J Alzheimers Dis 35, 751–760.

11.

Lopez-Sanz

, Bruna

, Garces

, Martin-Buro

, Walter

, Delgado

, Montenegro

, Lopez Higes

, Marcos

, Maestu

(2017) Functional connectivity disruption in subjective cognitive decline and mild cognitive impairment: A common pattern of alterations. Front Aging Neurosci 9, 109.

12.

Verfaillie

SCJ

, Binette

, Vachon-Presseau

, Tabrizi

, Savard

, Bellec

, Ossenkoppele

, Scheltens

, Flier

WMvd

, Breitner

JCS

, Villeneuve

, PREVENT-AD Research Group (2018) Subjective cognitive decline is associated with altered default mode network connectivity in individuals with a family history of Alzheimer’s disease. Biol Psychiatry Cogn Neurosci Neuroimaging 3, 463–472.

13.

Contreras

, Goñi

, Risacher

, Amico

, Yoder

, Dzemidzic

, West

, McDonald

, Farlow

, Sporns

, Saykin

(2017) Cognitive complaints in older adults at risk for Alzheimer’s disease are associated with altered resting-state networks. Alzheimers Dement 6, 40–49.

14.

Jack

Jr. , Bennett

, Blennow

, Carrillo

, Dunn

, Haeberlein

, Holtzman

, Jagust

, Jessen

, Karlawish

, Liu

, Molinuevo

, Montine

, Phelps

, Rankin

, Rowe

, Scheltens

, Siemers

, Snyder

, Sperling

, Contributors (2018) NIA-AA Research Framework: Toward a biological definition of Alzheimer’s disease. Alzheimers Dement 14, 535–562.

15.

Zelinski

, Gilewski

, Anthony-Bergstone

(1990) Memory Functioning Questionnaire: Concurrent validity with memory performance and self-reported memory failures. Psychol Aging 5, 388–399.

16.

Gilewski

, Zelinski

, Schaie

(1990) The Memory Functioning Questionnaire for assessment of memory complaints in adulthood and old age. Psychol Aging 5, 482–490.

17.

Merrill

, Siddarth

, Saito

, Ercoli

, Burggren

, Kepe

, Lavretsky

, Miller

, Kim

, Huang

, Bookheimer

, Barrio

, Small

(2012) Self-reported memory impairment and brain PET of amyloid and tau in middle-aged and older adults without dementia. Int Psychogeriatr 24, 1076–1084.

18.

Snitz

, Weissfeld

, Cohen

, Lopez

, Nebes

, Aizenstein

, McDade

, Price

, Mathis

, Klunk

(2015) Subjective cognitive complaints, personality and brain amyloid-beta in cognitively normal older adults. Am J Geriatr Psychiatry 23, 985–993.

19.

Amariglio

, Becker

, Carmasin

, Wadsworth

, Lorius

, Sullivan

, Maye

, Gidicsin

, Pepin

, Sperling

, Johnson

, Rentz

(2012) Subjective cognitive complaints and amyloid burden in cognitively normal older individuals. Neuropsychologia 50, 2880–2886.

20.

Nasreddine

, Phillips

, Bedirian

, Charbonneau

, Whitehead

, Collin

, Cummings

, Chertkow

(2005) The Montreal Cognitive Assessment, MoCA: A brief screening tool for mild cognitive impairment. J Am Geriatr Soc 53, 695–699.

21.

Sheikh

, Yesavage

(1986) Geriatric Depression Scale (GDS): Recent evidence and development of a shorter version. Clinical Gerontology 5, 165–173.

22.

Hill

, Mogle

, Wion

, Munoz

, DePasquale

, Yevchak

, Parisi

(2016) Subjective cognitive impairment and affective symptoms: A systematic review. Gerontologist 56, e109–e127.

23.

Zlatar

, Moore

, Palmer

, Thompson

, Jeste

(2014) Cognitive complaints correlate with depression rather than concurrent objective cognitive impairment in the successful aging evaluation baseline sample. J Geriatr Psychiatry Neurol 27, 181–187.

24.

Delis

, Kramer

, Kaplan

, Ober

(2000) California Verbal Learning Test–Second Edition (CVLT-II). The Psychological Corporation, San Antonio, TX.

25.

Silva

, Guerreiro

, Maroco

, Santana

, Rodrigues

, Bravo Marques

, de Mendonca

(2012) Comparison of four verbal memory tests for the diagnosis and predictive value of mild cognitive impairment. Dement Geriatr Cogn Dis Extra 2, 120–131.

26.

Donohue

, Sperling

, Salmon

, Rentz

, Raman

, Thomas

, Weiner

, Aisen

(2014) The preclinical Alzheimer cognitive composite: Measuring amyloid-related decline. JAMA Neurol 71, 961–970. https://surfer.nmr.mgh.harvard.edu/fswiki/FreeSurferMethodsCitation.

27.

FsFast, http://surfer.nmr.mgh.harvard.edu/fswiki/FsFast, Accessed June 10, 2020.

28.

Cox

(1996) AFNI: Software for analysis and visualization of functional magnetic resonance neuroimages. Comput Biomed Res 29, 162–173.

29.

Neuroimaging BmacotWCfH, Statistical Parametric Mapping, https://www.fil.ion.ucl.ac.uk/spm/, Accessed June 10, 2020.

30.

CONN: functional connectivity toolbox, https://www.nitrc.org/projects/conn, Accessed June 10, 2020.

31.

Artifact Detection Tools (ART), https://www.nitrc.org/projects/artifact_detect, Accessed June 10, 2020.

32.

Chu

PPW

, Golestani

, Kwinta

, Khatamian

, Chen

(2018) Characterizing the modulation of resting-state fMRI metrics by baseline physiology. Neuroimage 173, 72–87.

33.

Whitfield-Gabrieli

, Nieto-Castanon

(2012) Conn: A functional connectivity toolbox for correlated and anticorrelated brain networks. Brain Connect 2, 125–141.

34.

Burgmans

, van Boxtel

MPJ

, van den Berg

KEM

, Gronenschild

EHBM

, Jacobs

HIL

, Jolles

, Uylings

HBM

(2011) The posterior parahippocampal gyrus is preferentially affected in age-related memory decline. Neurobiol Aging 32, 1572–1578.

35.

Lazarou

, Nikolopoulos

, Dimitriadis

, Kompatsiaris

, Spilioti

, Tsolaki

(2019) Is brain connectome research the future frontier for subjective cognitive decline? A systematic review. Clin Neurophysiol 130, 1762–1780.

36.

Viviano

, Hayes

, Pruitt

, Fernandez

, van Rooden

, van der Grond

, Rombouts

, Damoiseaux

(2019) Aberrant memory system connectivity and working memory performance in subjective cognitive decline. Neuroimage 185, 556–564.

37.

Sorg

, Riedl

, Mühlau

, Calhoun

, Eichele

, Läer

, Drzezga

, Förstl

, Kurz

, Zimmer

, Wohlschläger

(2007) Selective changes of resting-state networks in individuals at risk for Alzheimer’s disease. Proc Natl Acad Sci U S A 104, 18760–18765.

38.

Vincent

, Snyder

, Fox

, Shannon

, Andrews

, Raichle

, Buckner

(2006) Coherent spontaneous activity identifies a hippocampal-parietal memory network. J Neurophysiol 96, 3517–3531.

39.

Kahn

, Andrews-Hanna

, Vincent

, Snyder

, Buckner

(2008) Distinct cortical anatomy linked to subregions of the medial temporal lobe revealed by intrinsic functional connectivity. J Neurophysiol 100, 129–139.

40.

Ward

, Schultz

, Huijbers

, Dijk

KRA

, Hedden

, Sperling

(2014) The parahippocampal gyrus links the default-mode cortical network with the medial temporal lobe memory system. Hum Brain Mapp 35, 1061–1073.

41.

Echávarri

, Aalten

, Uylings

HBM

, Jacobs

HIL

, Visser

, Gronenschild

EHBM

, Verhey

FRJ

, Burgmans

(2011) Atrophy in the parahippocampal gyrus as an early biomarker of Alzheimer’s disease. Brain Struct Funct 215, 265–271.

42.

Raji

, Lopez

, Kuller

, Carmichael

, Becker

(2009) Age, Alzheimer disease, and brain structure. Neurology 73, 1899–1905.

43.

Stoub

, deToledo-Morrell

, Stebbins

, Leurgans

, Bennett

, Shah

(2006) Hippocampal disconnection contributes to memory dysfunction in individuals at risk for Alzheimer’s disease. Proc Natl Acad Sci U S A 103, 10041–10045.

44.

Jeong

, Park

, Song

, Chung

, Rhie

(2017) Changes in cognitive function and brain glucose metabolism in elderly women with subjective memory impairment: A 24-month prospective pilot study. Acta Neurol Scand 135, 108–114.

45.

Loewenstein

, Barker

, Chang

J-Y

, Apicella

, Yoshii

, Kothari

, Levin

, Duara

(1989) Predominant left hemisphere metabolic dysfunction in dementia. Arch Neurol 46, 146–152.

46.

Rabin

, Smart

, Crane

, Amariglio

, Berman

, Boada

, Buckley

, Chetelat

, Dubois

, Ellis

, Gifford

, Jefferson

, Jessen

, Katz

, Lipton

, Luck

, Maruff

, Mielke

, Molinuevo

, Naeem

, Perrotin

, Petersen

, Rami

, Reisberg

, Rentz

, Riedel-Heller

, Risacher

, Rodriguez

, Sachdev

, Saykin

, Slavin

, Snitz

, Sperling

, Tandetnik

, van der Flier

, Wagner

, Wolfsgruber

, Sikkes

(2015) Subjective cognitive decline in older adults: An overview of self-report measures used across 19 international research studies. J Alzheimers Dis 48(Suppl 1), S63–86.

47.

Rock

, Roiser

, Riedel

, Blackwell

(2014) Cognitive impairment in depression: A systematic review and meta-analysis. Psychol Med 44, 2029–2040.

48.

Loprinzi

, Frith

(2018) The role of sex in memory function: Considerations and recommendations in the context of exercise. J Clin Med 7, 132.

49.

Figueroa

, Mocking

RJT

, van Wingen

, Martens

, Ruhé

, Schene

(2017) Aberrant default-mode network-hippocampus connectivity after sad memory-recall in remitted-depression. Soc Cogn Affect Neurosci 12, 1803–1813.

50.

Mah

, Szabuniewicz

, Fiocco

(2016) Can anxiety damage the brain? Curr Opin Psychiatry 29, 56–63.

51.

Mah

, Binns

, Steffens

, Alzheimer’s Disease Neuroimaging Initiative (2015) Anxiety symptoms in amnestic mild cognitive impairment are associated with medial temporal atrophy and predict conversion to Alzheimer disease. Am J Geriatr Psychiatry 23, 466–476.

52.

Yuen

, Rashidi-Ranjbar

, Verhoeff

, Kumar

, Gallagher

, Flint

, Herrmann

, Pollock

, Mulsant

, Rajji

, Voineskos

, Fischer

, Mah

(2019) Association between sleep disturbances and medial temporal lobe volume in older adults with mild cognitive impairment free of lifetime history of depression. J Alzheimers Dis 69, 413–421.

53.

Mah

, Anderson

, Verhoeff

NPLG

, Pollock

(2017) Negative emotional verbal memory biases in mild cognitive impairment and late-onset depression. Am J Geriatr Psychiatry 25, 1160–1170.